TO3 final P3A

[effebi]

Hi everybody! This is my first post here, thank you all for the hospitality!
Then I have from my job a lot of good MJ15003 / 4, couple of 500W 24-0-24 trafos, couple of 26MF 75 Mallory caps, bridges, etc. The idea is to build something simple, so my attention went on the nice P3A. I have scketched already a "split" version, where evething is on the PCB, except the two finals, their Emitter resistors, and the Zobel. The board sends the drive to the finals and gets back the output for the feedback. The power to the finals comes separately directly from the main caps, and on the bord I added a couple of 100uF. Everything else is original. Do you think is reasonable or am I going to fight hopelessly with auto - oscillations?
thanks for any advice.
effebi

[sakis]

obviously your schematic is wrong .... you need to double chek a few things ...

P3A is a CFP sziklai circuit ... it will oscilate eazier than a classic EFP ....

---to aviod things like that you need to minimize distance from transistors to pcb
--- then again it will be very wrong to use long cables to do that ( or twist B-C-E leads together )
--- so depenting on the way you heatsink the transistors it could be a possibility that you design a pcb that features can tranistors on board ...but trully this is heatsink depenting

The P3A is very well documented thanks to Rod ..... take my word for it each and every word written worth reading and will provide info to make perfect ...

post a few pictures of your potential heatsink ...then we will se if it can be done this way

read details regarding the vbe mutliplier and generally read the article carefully .....

some more tips can be found here P3A COMPARSION TABLE ( long .... )

and after a few minutes check your PM list i will send you an idea ...

kind regards sakis

[effebi]

obviously your schematic is wrong .... you need to double chek a few things ...

If you mean I should bring out collectors, yeah, right :-)

P3A is a CFP sziklai circuit ... it will oscilate eazier than a classic EFP ....

This is my main concern...

read details regarding the vbe mutliplier and generally read the article carefully .....

If you mean that I used a BD139, I did it for hanging it on a small Heatsink with the drivers (NOT the main Heatsink), in order to thermal compensate. I supposed the smaller hfe should not harm.

Thank you
effebi

[jaycee]

It can work. I did quickly build one with MJ15003/4 myself and it worked OK. Keep the wires to the output transistors short, no more than 2 inches. Put 220uF decoupling capacitors on your PCB, near to the driver transistors. Use the BD140 as a VAS rather than an MJE15035 as it is more suitable. It'll be ok with 35v supplies, but fit a heatsink to it to be sure.

One thing I found essential for the P3A - ground the heatsink!

[effebi]

Use the BD140 as a VAS rather than an MJE15035 as it is more suitable

Culd you please elaborate? from the author article looks like the new MJs are BETTER, and the data sheet hints to me the same. Any specific parameter?
thanks
effebi

[effebi]

So I have made up revised schematic, taking in accont the suggestions from the Forum and the improvements done by the Author in P68.
In particular there are parallel capacitors for high frequency in the input and feedback, ground has been separated for input and power, and a resistor was added on the collector of Q2. Power supply is decoupled with additional 100uF per rail. The only "new" thing is VBE transistor being a TO220 one like drivers, in order to hang them all together in some small heat sink. I have so chosen to keep the MJ15034/5 as drivers, since with the very large SOA of the MJ15003/4 final I might try to risk to work with some low-ohm load at +-40 V rails.

http://www.atgraf.com/P3A-68_1.pdf
Any comment before I am going to buy the few missing components?
Thank you all in advance.
effebi
P.S. a suitable PCB was specially designed to suit the TO3 and existing heat sink, but, since I understood the author does not like that, it will be not shown here.

[effebi]

I have also added some Ic currrent compensation on the input LTP. Some people on this Forum say unbalanced currents may produce noticeable distorsion. Also the offset might be affected. BTW I am planning to use 2n5550 on that part. Do you think it is ok? Or just with mathched components and leaving the feedback leg without collector resistor is enough?
Thanks
effebi